Little PICs with hi-res A-D ?

[Eeyore]

Scott Seidman wrote:

Eeyore wrote
Lostgallifreyan wrote:
Eeyore wrote

You might be wondering why but have you ever heard of a
phenomenon called 'zipper noise' ?

Can't integrate the step response on the output?

It's not driving a VCA.

Something similar was done on the control signals of Yamaha gear.
Worked fine except they overdid it and made it sluggish.

Yes quite ! That's no good here, needs to be 'real time'.


Unless you also notice discrete steps when not actually
stepping the resolution might be fine, just needing smoothing of
changes.

This needs to be used for actual mixing so the steps need to be tiny.

There's several possible solutions including full DSP but that's more
work.

Don't use pots for input-- use encoders. Then the resolution of each step
is whatever you program it to be. You can even program speed
sensitivity, or course and fine adjust.

The step size on encoders is even worse !

Graham

 

[Eeyore]

Hal Murray wrote:

You might be wondering why but have you ever heard of a
phenomenon called 'zipper noise' ?

Unless you also notice discrete steps when not actually
stepping the resolution might be fine, just needing smoothing of changes.

This needs to be used for actual mixing so the steps need to be tiny.

Can you hear the difference between steps at 8 bits?

In excruciating detail. We are talking about professional CD/record production
here. Sounds like the youngster made the same assumptions as you.

Graham

 

[Lostgallifreyan]

Eeyore <rabbitsfriendsandrelations@hotmail.com> wrote in
news:48F79AC3.F1EB31DF@hotmail.com:

and be fast enough for real time. How fast CAN a mix engineer move a
knob?

Very fast.


I have just suggested an analogue solution in fact.

Not as fast as a transient capable of an audible click. Not even if he
accelerated a lever that then impacted it up the scale. There's a window of
timing range there that you can use. Good to hear an analog answer works
though. People didn't work for years at brilliant answers so we can forget
them the moment digital models can be made. Some things just aren't that
easily modelled, and it interests me that when I asked the same question you
did (about PIC's with high bit depth conversion), for different purposes, I
also got several suggestions that don't really hit the spot. A year later,
this is still true, yet FORTY years ago, my problem was solved by Bob Moog
and others with a couple of op-amps and a pair of matched transistors. Which
is how I'll do it too, if I ever get back to that project...

 

[Scott Seidman]

Eeyore <rabbitsfriendsandrelations@hotmail.com> wrote in
news:48F79B00.111E6E2F@hotmail.com:

The step size on encoders is even worse !

The step size on encoders is 1. What you do to it after you collect it is
your business. If you multiply it by 2, then its 2. If you divide it by
100, its 0.01. My point is that for an input device in a digital system,
resolution doesn't have to be a factor.

--
Scott
Reverse name to reply

 

[Scott Seidman]

Scott Seidman <namdiesttocs@mindspring.com> wrote in
news:Xns9B39ADC8A8E4Ascottseidmanmindspri@130.133.1.4:

Eeyore <rabbitsfriendsandrelations@hotmail.com> wrote in
news:48F79B00.111E6E2F@hotmail.com:

The step size on encoders is even worse !

The step size on encoders is 1. What you do to it after you collect it
is your business. If you multiply it by 2, then its 2. If you divide
it by 100, its 0.01. My point is that for an input device in a
digital system, resolution doesn't have to be a factor.

For example, I'm not sure any of my scopes have any mechanical pots on them
anymore.

 

[donald]

Scott Seidman wrote:

For example, I'm not sure any of my scopes have any mechanical pots on them
anymore.

All kinds of scopes are going digital:
http://www.opticsbestbuy.com/ELCAN-DigitalHunter-DayNight-Digital-Riflescope-ELDHDN.html

donald

 

[Eeyore]

Scott Seidman wrote:

Eeyore wrote

The step size on encoders is even worse !

The step size on encoders is 1. What you do to it after you collect it is
your business. If you multiply it by 2, then its 2. If you divide it by
100, its 0.01. My point is that for an input device in a digital system,
resolution doesn't have to be a factor.

Mixing engineers don't want controls that need turning 3000 degrees. 300 is
what they want.

Graham

 

[Eeyore]

Scott Seidman wrote:

Scott Seidman <namdiesttocs@mindspring.com> wrote
Eeyore <rabbitsfriendsandrelations@hotmail.com> wrote

The step size on encoders is even worse !

The step size on encoders is 1. What you do to it after you collect it
is your business. If you multiply it by 2, then its 2. If you divide
it by 100, its 0.01. My point is that for an input device in a
digital system, resolution doesn't have to be a factor.

For example, I'm not sure any of my scopes have any mechanical pots on them
anymore.

Quite possibly but you don't mix on them do you ?

Graham

 

[Lostgallifreyan]

Scott Seidman <namdiesttocs@mindspring.com> wrote in
news:Xns9B39ADC8A8E4Ascottseidmanmindspri@130.133.1.4:

The step size on encoders is even worse !

The step size on encoders is 1. What you do to it after you collect it is
your business. If you multiply it by 2, then its 2. If you divide it by
100, its 0.01. My point is that for an input device in a digital system,
resolution doesn't have to be a factor.

Physical step angle does though.

 

[Scott Seidman]

Eeyore <rabbitsfriendsandrelations@hotmail.com> wrote in
news:48F7BA4D.97A539DB@hotmail.com:

Mixing engineers don't want controls that need turning 3000 degrees.
300 is what they want.

Graham

I think you'd be quite suprised about how easy the transition between pot
and frob knob is. You can make a frob knob's sensitivity speed sensitive
if you program it that way. You can store a whole bunch of presets,
which you couldn't dream of with pots. You can get them with 2^10 counts
per revolution, which is 10 full bits, and you never need to worry about
noise, or anything like it, or kid yourself into thinking that you have
12 useful bits because you have a 12 bit ADC.

http://lgrws01.grayhill.com/web/images/ProductImages/Opt_Encoder_61K_
61R.pdf (1 "cycle" is 4 counts)

It might seem expensive, but you might not need as many of them as you
would with pots.

A bit more research is showing me 2048 counts!

http://www.secomtel.com.cn/UpFilesPDF/PDF/Agilent/PDF_DOCS/ISONCONT/02
_MOTN/2_116_18.pdf

Free up your design a bit. Just because an analog mixing board has a
pots controlling amps, that doesn't mean that you need a pot on your
input. You could even get a little imaginative with the mechanics. A
mechanical mouse uses low res encoders, but they're effectively geared to
give killer resolution.



--
Scott
Reverse name to reply

 

[Eeyore]

Scott Seidman wrote:

Eeyore wrote

Mixing engineers don't want controls that need turning 3000 degrees.
300 is what they want.

I think you'd be quite suprised about how easy the transition between pot
and frob knob is. You can make a frob knob's sensitivity speed sensitive
if you program it that way.

That's not the way mixing engineers work. They need an absolute position to
correspond to an absolute volume/gain.

Graham

 

[Jasen Betts]

On 2008-10-14, Eeyore <rabbitsfriendsandrelations@hotmail.com> wrote:

I guess you could try oversampling. If you really need more than 10 bits,
maybe an external chip? You really haven't provided much information on
your ADC performance requirements.

Doesn't need to be fast for one if that's what you mean. It's for scanning
some level controls on a pro-audio product. That's why we want better than 10
bit resolution. You might be wondering why but have you ever heard of a
phenomenon called 'zipper noise' ?

could you interpolate in software ?
response will lag behind manipulation but if done right not by much.

Bye.
Jasen

 

[Robert Lacoste]

"Eeyore" <rabbitsfriendsandrelations@hotmail.com> a écrit dans le message de
news: 48F45665.A5C64205@hotmail.com...

Me ? A PIC ? You must think I'm going mad !

No, it's for someone else I may work on a project with and he's into
PICs. He wants a little 8 pin jobbie with an onboard A-D. Currently he's
looking at a 16F220 IIRC. That only has an 8 bit A-D.

I reckon we need 12 bits off the top of my head but 16 would be nice.

Any suggestions ? I'm totally unfamiliar with their range and he's off
on dog-sitting holiday for 2 weeks.

Why not a PSoC from Cypress ? You can implement a 14 bit ADC in its analog
configurable blocs, and there are 8-pin devices in the family.

Yours,

--
Robert Lacoste
ALCIOM - The mixed signal experts
www.alciom.com

 

[Eeyore]

Robert Lacoste wrote:

"Eeyore" a écrit

Me ? A PIC ? You must think I'm going mad !

No, it's for someone else I may work on a project with and he's into
PICs. He wants a little 8 pin jobbie with an onboard A-D. Currently he's
looking at a 16F220 IIRC. That only has an 8 bit A-D.

I reckon we need 12 bits off the top of my head but 16 would be nice.

Any suggestions ? I'm totally unfamiliar with their range and he's off
on dog-sitting holiday for 2 weeks.

Why not a PSoC from Cypress ? You can implement a 14 bit ADC in its analog
configurable blocs, and there are 8-pin devices in the family.

You mean semi-custom ?

We don't have those quantities.

Graham

 

[Eeyore]

Jasen Betts wrote:

Eeyore <rabbitsfriendsandrelations@hotmail.com> wrote:

I guess you could try oversampling. If you really need more than 10 bits,
maybe an external chip? You really haven't provided much information on
your ADC performance requirements.

Doesn't need to be fast for one if that's what you mean. It's for scanning
some level controls on a pro-audio product. That's why we want better than 10
bit resolution. You might be wondering why but have you ever heard of a
phenomenon called 'zipper noise' ?

could you interpolate in software ?
response will lag behind manipulation but if done right not by much.

Well, other factors are now looking like pushing the project back onto a more
conventional route besides NXP has IIRC 12 bit ADCs in its slimmed down 8051
derived LPC700 and 900 ranges. Sod the PICs I'm an 8051 expert apparently (or so I
was once told).

Graham

 

[Eeyore]

kevin93 wrote:

Eeyore wrote:
...
LOL @ 2.7V. They really don't want to make life easy for you these days do
they ?
Oh well, it'll interface to a 10F220 at least ! Just hope everything else
likes 2.7V !
..
Graham

2.7V is the lower limit - it supports 2.7-5.5V

Typical Microchip data sheet. Didn't make it clear AND I looked.

Graham

 

[Eeyore]

Richard Henry wrote:

Eeyore wrote:
Me ? A PIC ? You must think I'm going mad !

No, it's for someone else I may work on a project with and he's into
PICs. He wants a little 8 pin jobbie with an onboard A-D. Currently he's
looking at a 16F220 IIRC. That only has an 8 bit A-D.

I reckon we need 12 bits off the top of my head but 16 would be nice.

Any suggestions ? I'm totally unfamiliar with their range and he's off
on dog-sitting holiday for 2 weeks.

What is the magnitude (in bits) of the noise in the system (for
whatever you want to call "noise")?

Sod all noise. Effectively clean DC. Just want the resolution. Doesn't even
have to fast. Although this plan is now beginning to look shaky (for entirely
other reasons) and plan B or C may be used instead.

Graham

 

[kevin93]

On Oct 14, 6:22 am, Eeyore <rabbitsfriendsandrelati...@hotmail.com>
wrote:
....

LOL @ 2.7V. They really don't want to make life easy for you these days do
they ?
Oh well, it'll interface to a 10F220 at least ! Just hope everything else
likes 2.7V !
...
Graham

2.7V is the lower limit - it supports 2.7-5.5V

kevin

 

[Richard Henry]

On Oct 14, 1:20 am, Eeyore <rabbitsfriendsandrelati...@hotmail.com>
wrote:

Me ? A PIC ? You must think I'm going mad !

No, it's for someone else I may work on a project with and he's into
PICs. He wants a little 8 pin jobbie with an onboard A-D. Currently he's
looking at a 16F220 IIRC. That only has an 8 bit A-D.

I reckon we need 12 bits off the top of my head but 16 would be nice.

Any suggestions ? I'm totally unfamiliar with their range and he's off
on dog-sitting holiday for 2 weeks.

Graham

What is the magnitude (in bits) of the noise in the system (for
whatever you want to call "noise")?

 

[Andre Majorel]

On 2008-10-14, Eeyore <rabbitsfriendsandrelations@hotmail.com> wrote:

Anthony Fremont wrote:
Eeyore wrote:
Anthony Fremont wrote:
David L. Jones wrote:

12F683 is the pick of the bunch:
http://www.microchip.com/wwwproducts/Devices.aspx?dDocName=en010115
only 10bit ADC, Microchip haven't figured out how to do
12 bit yet.

I agree, that's the best 8 pin PIC I know of. Don't be
fooled by the 12F, it's still a 14 bit core like the 16F
series.

How about a couple with 'windowing' ? Can that be done to
extend the resolution ?

I guess you could try oversampling. If you really need more
than 10 bits, maybe an external chip? You really haven't
provided much information on your ADC performance
requirements.

Doesn't need to be fast for one if that's what you mean. It's
for scanning some level controls on a pro-audio product.
That's why we want better than 10 bit resolution.

Don't know what kind of control this is but IME, a rotary pot
with a 15 mm knob can be reliably set to about 1/350th of the
range. That's less than 9 bits.

You might be wondering why but have you ever heard of a
phenomenon called 'zipper noise' ?

Wouldn't that be better addressed by some averaging and
interpolation in software ? That would get rid of both zipper
noise and jitter.

--
André Majorel <URL:http://www.teaser.fr/~amajorel/>
You measure democracy by the freedom it gives its dissidents, not
the freedom it gives its assimilated conformists -- Abbie Hoffman.